Title :
Global word shape processing in off-line recognition of handwriting
Author :
Parisse, Christophe
Author_Institution :
Inst. Nat. de la Sante et de la Recherche Med., Paris, France
fDate :
4/1/1996 12:00:00 AM
Abstract :
Off-line recognition of handwriting may be achieved using simplified profiles of word shapes. These profiles consist of approximations of the word´s upper and lower contour. Training and recognition are based on n-gram extraction and identification. The lexicons used extend to 16,000 words
Keywords :
handwriting recognition; learning (artificial intelligence); global word shape processing; handwriting; identification; lower contour; n-gram extraction; off-line recognition; upper contour; word shapes; Accidents; Dynamic programming; Electronic mail; Handwriting recognition; Humans; Image recognition; Pattern recognition; Prototypes; Shape; Stability analysis;
Journal_Title :
Pattern Analysis and Machine Intelligence, IEEE Transactions on